Dirichlet Process Priors for Small Area Estimation and Disease Mapping


Ranking, Sparseness and Clustering in Disease Mapping

Malay Ghosh (Department of Statistics, University of Florida) (Speaker)
Xueying Tang (Department of Statistics, University of Florida)

We illustrate some applications of Dirichlet Process Priors for modeling random effects in small area models. We also show some of its application in disease mapping by using a Dirichlet process prior with a baseline CAR model. One advantage of such priors is that there is automatic clustering as well as tracking a multimodal posterior much more accurately than normal random effects model.